3.1.1 ‘Post-verbal’ Ideophones, Part One: Or Why Ideophones Reduplicate

Numerous descriptions of ideophones in the languages where they occur mention that they are frequently repeated or either fully or partially reduplicated (Paul Reference Paul1880:181–2; Doke Reference Doke1935:118; Samarin Reference Samarin1965, Reference Samarin1971; Wescott Reference Wescott1973, Bakker Reference Bakker and Kouwenberg2003:75 among very many: see Table 3.1), as indeed is tsktsk, or the negative grunt commonly transcribed as unhunh. The tendency of ideophones to be repeated has only been identified as criterial by one linguist (Lee) working on one language (Korean), and is not a universal definitional property (cf. the English ideophones yuk, duh, phooey), nor has it ever been explicitly proposed as one, although Childs (Reference Childs1989, Reference Childs and Hinton1994), Lee (Reference Lee1992), Sohn (Reference Sohn1994:496), Asher and Kumar (Reference Asher and Kumar1997:447), Diffloth (Reference Diffloth and Jenner1976:251), Beck (pdf), and Enfield (Reference Enfield2007:299) all do come pretty close. The question is, why should they reduplicate at all?

The fact that they do so is incompatible with the hypothesis that they are missing links to some prelinguistic utterances. Not only is there nothing prelinguistic or primitive a priori about repetition or reduplication, in which case one of the most striking properties of ideophones is unexplained by the missing link hypothesis. In fact, there is evidence from numerous languages that ideophones are actually formed via reduplication from prosaic nouns, verbs, and adjectives. I will argue however that repetition and its grammaticalized derivative reduplication can provide evidence for ideophones as a missing link between gestures and words.

Table 3.1 lists some of the languages in which reduplication is prominently mentioned as a feature of ideophones.

Indeed, if we agree that grammatical reduplication, whether of entire words or of parts, is nothing but conventionalized repetition (cf. Gil Reference Gil and Hurch2005), this tendency may be even greater than the grammarians who have described ideophones are willing to explicitly acknowledge. For example, only about half of the seventeen ideophones listed in Olawsky’s grammar of Urarina (an isolate of Peru) are identified as reduplications, and njaaw “expression of lust” is not one of them. In a short text passage featuring this word, however, it appears repeated three times the first time it appears, and two times the second (Olawsky Reference Olawsky2006:842):

Ideophones, that is to say, may not always be (obligatorily) reduplicated: but, with the exception of ideophones like bang, which are specifically produced to imitate one-time events, they are (optionally) repeated, with the number of repetitions being a matter of individual style. Huttar and Huttar (Reference Huttar and Huttar1994:602) point out that in Ndyuka, where only thirty-nine of the ninety-seven ideophones that they list do in fact reduplicate, the capacity of ideophones to reduplicate four or more times is one of the phonologically odd features that distinguish them from other parts of speech. In Kisi (Childs Reference Childs and Hinton1994:136), in Khwe (Kilian-Hatz Reference Kilian-Hatz2008:245), and Nyulnyul and Gooniyandi (McGregor Reference McGregor, Voeltz and Kilian-Hatz2001:210), reduplication is the only morphological process that ideophones undergo. Foris (Reference Foris2000:40) notes that while not all ideophones reduplicate in Chinantec, many of them may. Kruspe (Reference Kruspe2004:402) points out that in Semelai, ideophonic reduplication differs from that of other manner adverbs in that ideophones undergo total reduplication of the whole root, rather than simply repetition of a single consonant or a light syllable.

The most commonly cited motivation for reduplication in general is iconic: repetition directly signals plurality, intensivity, iteration, or duration (cf. Sapir Reference Sapir1921:76). And the prevailing opinion among authors is to ascribe ideophonic reduplication entirely to iconicity: that is, the repetition of the word simply denotes intensity or repetition of the action or sense datum it stands for (e.g. Diffloth Reference Diffloth and Jenner1976:252 and passim for Semai; Sohn Reference Sohn1994:501 for Korean; Kita Reference Kita1997:400; Hamano Reference Hamano1998:65, 104, for Japanese; Klamer Reference Klamer1998:245 for Kambera; Schultze-Berndt Reference Schultze-Berndt, Voeltz and Kilian-Hatz2001:359 for Jaminjung; Rubino Reference Rubino, Voeltz and Kilian-Hatz2001:309 for Ilocano; Noss Reference Noss, Voeltz and Kilian-Hatz2001:262 for Gbaya; Kilian-Hatz Reference Kilian-Hatz, Voeltz and Kilian-Hatz2001:158 for Baka and 2008:245 for Khwe; de Jong Reference De Jong, Voeltz and Kilian-Hatz2001:127 for Didinga; Bakker Reference Bakker and Kouwenberg2003:76 for Saramaccan; Kruspe Reference Kruspe2004:402 for Semelai; Epps Reference Epps2008:721–4 for Hup; Beck pdf for Upper Necaxa Totonac; Wälchli Reference Wälchli and Arkadiev2015 for Lithuanian).

While this iconic motivation is certainly true of some possible ideophones, like snickersnack or zigzag, it is not true of all of them (recall unhunh), and at least one of the authors in Voeltz and Kilian-Hatz’s anthology explicitly denies that “expressive” (or iconic) reduplication and ideophonic reduplication are the same thing (Roulon-Doko Reference Roulon-Doko, Voeltz and Kilian-Hatz2001:291). Others, while ascribing high iconicity to each of the members of a rhyming or alliterative pair, are silent on whether the repetition itself performs any iconic function (Ratliff Reference Ratliff1992). I side with Roulon-Doko in questioning my sources. While ideophonic repetition may often signal some kind of multiplication, there are other cases where it seems to me that it cannot. Below are some randomly culled examples of ideophones whose reduplication seems to carry little or no hint of any iconic motivation:

1. 1. (Cilubà: Kabuta Reference Kabuta, Voeltz and Kilian-Hatz2001:142,143, 150)

   a) kabakaba	embarrassment
   b) zeezeeze	whiteness
   c) myamyamya	scarcity
   d) bukonkoto	avarice

2. 2. (Emai: Egbokhare Reference Egbokhare, Voeltz and Kilian-Hatz2001:88)

   a) ririri	red
   b) woziwozi	obese

3. 3. (Wolaitta: Amha Reference Amha, Voeltz and Kilian-Hatz2001:50, 51)

   a) yook’I yook’a	sympathetic, emotionally weak
   b) puski puska	agile

4. 4. (Yoruba: Pulleyblank Reference Pulleyblank and Comrie1990:981)

   a) rokiroki	red
   b) rodorodo	bright
   c) rubuturubutu	round
   d) potopoto	soft mud
   e) jalajalajalajala	shabby

5. 5. (White Hmong: Ratliff Reference Ratliff1992:218–45, passim)

   a) mplihb mpleeb	pin coming out of hand grenade
   b) mlig mlog	tiger calling
   c) pligplog	body jumping into water
   d) mij mewj	[NB a single] bullet whizzing past
   e) quj qaim	slim, svelte

6. 6. Kisi (Childs Reference Childs and Hinton1994:186)

   a) kpiini kpiini	foul smelling
   b) dong dong	quietly
   c) cam cam	lukewarm

7. 7. Lao (Enfield Reference Enfield2007:299–303)

   a) qeet5 teet5	completely vacant or deserted
   b) ciim1piim1	of people or things, faintly visible in the distance
   c) kuu1 suu1	head bowed
   d) ciing1 piing1	tiny hole (too small to put a finger in)
   e) cùùng1 khuung1	deep, pure red

8. 8. Siwu (Dingemanse Reference Dingemanse2009a, Reference Dingemanseb)

   a) (kpina)4	black
   b) (gele)4	shiny
   c) gbègbèègbè	stiff
   d) kpokporo	going strong, in glowing health
   e) (tsin)5	neatly
   f) ka(na)7	silent

9. 9. Udihe (Nikoleva and Tolskaja Reference Nikolaeva and Tolskaja2001:381–3)

   a) (kälj)2	pale
   b) (zile)2	stripy
   c) (bugdumce)2	slippery
   d) (cifu)2	wet
   e) (kiᶇde)2	break
   f) (ciᶇu)2	swing

10. 10. Sgaw Karen (Ratanakul pdf:131–2)

    a) k ǝ?1swe2 kǝ?1 swɔ2	describing soil that can let water pass freely
    b) ( kǝ 1bɛ’ 3)2	of actions that are slow
    c) pǝ’ci 1 pǝ’ te2	(to wait/to see with) impatience
    d) (bla2)2	to have no taste such as in badly prepared soup

11. 11. Sre (Diffloth Reference Diffloth and Thongkum1979:53–5)

    a) pɔh-smpɔh	seriously, not halfway
    b) ŋuu-rә-ŋuu	livid
    c) ‘ɛɛw-rә-’ɛɛw	feeling of weakness
    d) tәәt –mhәәt mhәәt	appearance of a very flat nose
    e) tun-plun plun	floppy, of (a dog’s) dangling ears

Non-iconic reduplication in such cases can be thought of as ‘non-referential’, to choose a neutral word for the time being. But ‘non-referential’ includes a variety of possibilities and we can still distinguish between purely grammatical (=unmotivated) and otherwise motivated repetition. Purely grammatical and arbitrary reduplication, attested in Greek, Latin, and Gothic, for example, creates the perfect tense/aspect: parco: ‘I spare’ versus pe-perci ‘I have spared’. Other kinds of motivation, non-referential in nature, may exist for reduplication: playfulness, irony and even counter-iconic diminution, where the string {original + copy} signals a weaker version of what is signaled by the original alone (Pilhofer Reference Pilhofer1933:85; Moravcsik Reference Moravcsik and Greenberg1978:323; Thompson Reference Thompson1987:172–3).

One of these motivations may be the love of symmetry for its own sake, as in English decorative compound expressions like true-blue, loosey-goosey, legal beagle, spic ‘n’ span. The syntagm {original + copy} seems to mean exactly the same thing as the original by itself. This is a controversial claim, and in most accounts of Western languages forms like these frequently suggest ‘disordered iteration’ (cf. Wälchli Reference Wälchli2005) but the purely decorative function of such reiteration seems to be widely accepted in accounts of various Southeast Asian languages (see Nacaskul Reference Nacaskul and Jenner1976 for an early account). The symmetry may be between one meaningful word and another, which, though meaningful elsewhere, has apparently been coopted to serve as a partner for the main one on the basis of its sound alone. Such pairs shade into others, where either one (e.g. jibber jabber) or both (e.g. helter skelter, pell mell, hurly burly, hanky panky, and the like) of the partners in the symmetrical syntagm are currently meaningless. Such pairs are known by various names, and I will adopt Marchand’s (Reference Marchand1960) characterization of them as “twin forms.”

It is presumably through the widespread incidence of apparently decorative symmetry more than any other single feature that ideophones have been recurrently identified with ‘twin forms’ (Paul Reference Paul1880; Westermann Reference Westermann and Meinhof1927:321; Smithers Reference Smithers1954; Diffloth Reference Diffloth1972; Wälchli Reference Wälchli2005:chapter 5). In particular, both productions seem to use repetition metalinguistically to draw attention to themselves (Samarin Reference Samarin1971:165; Dingemanse Reference Dingemanse2011:70).

I do not myself regard twin forms as ideophones, but will defer fuller discussion of the formal and semantic differences between them until the final chapter of this book.

Another motivation for repetition is infantile: Ferguson has claimed that reduplication is a recurrent if not universal feature of baby talk in different languages (Ferguson Reference Ferguson1964), and it is well-known that child language, or motherese, simplifies bisyllabics in the direction of reduplication (gateau > tato), and reduplicates monosyllabics to conform to the same syllabic template (bon > bon bon).

And yet another is the expression of pejorative dismissal. English has borrowed the word-formation device of ‘schmo reduplication’ (e.g. justice schmustice), variations of which occur in several other languages, among them Turkish and Vietnamese: cf. Thompson (Reference Thompson1987:173) for ‘ironic emphatics’ ending I –iêc in Vietnamese, Lewis (Reference Lewis1967:237) for ‘m-doublets’ in Turkish. Stolz (Reference Stolz, Siemund and Kintana2008) provides a cross-linguistic survey.

I would like to introduce here yet another kind of possible motivation for reduplication. It may be histrionic: The essence of a performance or showing off in general is that it is overdone, and one highly conventionalized kind of overdoing anything may be repetition. (The locus classicus for repetition as a kind of ‘showing off’ in American literature is the thus-titled chapter in Tom Sawyer, although the practice is actually better exemplified in a passage from Heller’s Catch-22, where Colonel Korn is ostensibly briefing his men for a flight over Avignon, but really parading his competence and leadership for a visiting senior officer (Heller Reference Heller1972:229).)

And there may be many other motivations. Which is likely to be the crucial factor in ideophonic reduplication?

We can eliminate infantile motivation in principle. Motherese, a traditional source for reduplicating expressions like googoo, dada, may owe this aspect of its structure not to the mothers (as plausibly held by Paul Reference Paul1880), but surprisingly, to the children themselves (cf. Oller 1978; Fee and Ingram 1980; Ferguson 1983; Stoel-Gammon and Otomo 1986). This would recapitulate nicely the sense that there is something not only primitive, but downright infantile, about twin forms to many people, and may go some way toward accounting for the snobbism with which they and ideophones are so often disregarded. But there is no reason to suppose that adult speakers using ideophones are going out of their way to sound babyish.

There is suggestive rather than technical evidence, from languages of Southeast Asia, that iconic reduplication and the other main kinds of non-referential reduplication (whether decorative or histrionic) are grammatically different: Decorative reduplication is not subject to phonetic reduction, presumably because reduction would destroy the formal symmetry that is its reason for existence. Iconic reduplication is thus reduced. Although I will return to this distinction in Chapter 7, here is a brief illustration by way of a preview, from Wa (Watkins Reference Watkins and Williams2014).

Iconic reduplication undergoes reduction to create an iambic lexeme:

Decorative reduplication, however, does not undergo reduction, and symmetry is preserved:

This leaves decorative and histrionic motivation as plausible contenders for what is going on in SE Asian words of the jibber jabber type. I can demonstrate on purely formal grounds that they do not behave like ideophones in at least one SE Asian language, Khmer. When they are reduplicated at all (which is most of the time), ideophones almost invariably consist of faithful total repetitions of a root:

Decorative compounds, on the other hand, a favorite lexical construction in the language, consist of a root accompanied by a (usually meaningless, and therefore unglossed) alliterating ‘companion word’, which derives from a number of conjectural sources:

Ideophones, then, are typically reduplications, but they are – at least in some languages – formally distinct from iconically or decoratively reduplicated words. I maintain that they reduplicate not only in Khmer, but so often in other languages as well, not (or not only) because they often signal plurality, intensification, or repetition, and not because they are decorative, but because they are uttered as conventionalized performances. The functional unity of ‘expressive lengthening’ – itself a histrionic display – and reduplication in ideophones is underlined by several observers, among them Smithers (Reference Smithers1954), Newman (Reference Newman, Voeltz and Kilian-Hatz2001), Egbokhare (Reference Egbokhare, Voeltz and Kilian-Hatz2001), Epps (2005), Heath (ms.), and Dingemanse, cited in extenso on pages 90-92 of Chapter 2. Dingemanse observes (Reference Dingemanse2009a:1) that in Siwu ideophones are “morphologically special because they freely undergo various types of reduplication and lengthening” (my emphasis). Examples of the latter include:

He goes on to say (Reference Dingemanse2009b:2) that ID are “performatively foregrounded, that is to say, they freely undergo various types of reduplication, lengthening, and expressive intonation” – as prototypically histrionic performances may do.

In extreme cases, this functional unity of histrionic deformation and reduplication even seems to be formally expressed by their actual complementary distribution in some languages, with expressive lengthening applying to the final vowel or consonant of an ideophone, or repetition to its final syllable (Maduka Reference Maduka1988, passim; Roulon-Doko Reference Roulon-Doko, Voeltz and Kilian-Hatz2001:291; Noss Reference Noss, Voeltz and Kilian-Hatz2001:262). That is, there are languages in which ideophones may apparently undergo either expressive lengthening or reduplication. Thus, in Gbaya

So, it may be that reduplication of ideophones is compatible with the missing link hypothesis after all if the original motivation is histrionic.

But what can we make of the frequently repeated observation that ideophones are actually created by repetition or reduplication? In the most important paper on the etymology of ideophones that I have seen, Childs flatly asserts that most ideophones in Kisi typically originate simply as reduplications of verbs (Reference Childs1989:55–6, 58). Some of his examples follow:

In general, a Kisi verb is reduplicated for emphasis. For greater emphasis, the second part may be formally altered, either prosodically, with extra-high tones, or segmentally, for example, by changing a vowel. The resulting form is often an ideophone in Kisi.

Reduplication can also, less regularly, produce ideophones from nouns.

Following Childs, Brunelle and Xuyen assert that reduplication alone may be what creates ideophones from non-expressive verbs in Vietnamese (Reference Brunelle, Xuyen and Williams2014:85).

In Khwe, coe is the apparently prosaic name of the black-cheeked lovebird, but the reduplication of the name is an ideophone representing its call (Kilian-Hatz Reference Kilian-Hatz2008:246).

In Hausa, reduplication is one of three means whereby ideophones may be derived from prosaic words: the other two are loss of a stem-final vowel, and expressive deformation of the length or tone of the final syllable (Newman Reference Newman, Voeltz and Kilian-Hatz2001:255). Thus caka ‘stab’ is related to the ideophone cako-cako ‘sharp, pointed’; ɗ’age ‘shrink’ is related to the ideophone ɗ’age -ɗ’age ‘skimpy’.

In Siwu, nouns and adjectives may be reduplicated to form ideophones (Dingemanse Reference Dingemanse2011:153). Thus, the noun ɔmɛrɛ ‘sweetness’ may produce the ideophone mɛrɛ mɛrɛ ‘sweet’; bɛ ‘wide’ is the certain source of the ideophone bɛbɛɛbɛ; mini ‘encircle’ is the certain source of minimini ‘round’.

In Udihe, bugdu means ‘slippery’, and is an adjective, but (bugdumce)2 also meaning ‘slippery’, is an ideophone (Nikolaeva and Tolskaja Reference Nikolaeva and Tolskaja2001:383).

In Semai, where “as a rule color terms and words for taste are ambivalently expressive and stative verbs,” it is reduplication that separates the ideophonic sheep from the verbal goats The root cnga:l “red” is a case in point. Reduplicated, it is the ideophone ci-cgna:l “appearance of a flickering red object,” as a verb, it yields a conventional series of derivations via other affixes, such as c-r-nga:l “redden” (causative) (Diffloth Reference Diffloth and Jenner1976:255).

In Korean, not only are the vast majority of inherent ideophones reduplicated. Words that are part of the normal prosaic vocabulary are made ideophonic through reduplication. Examples include (cakum)2 “in small bits” from cak-ta “small”, (etwuk)2 “darkish” from etwup-ta “dark”, and (nokul)2 “soft, tender” from nok-ta “melt” (Sohn Reference Sohn1994:496), (hanɯl)2 “graceful, airy”, from hanɯl ‘sky’ (DJ Kim, p.c.).

Finally, it may be that reduplication is one of the means whereby ideophones are derived from prosaic verbs in Lithuanian. The conventional verb žvalg ~ žvelg ‘look around’ may appear with the ideophone (or intensifier? or figura etymologica?) žvilgt žvilgt in

“He looked around.” (Wälchli Reference Wälchli and Arkadiev2015)

This harnessing of reduplication as a derivational means for producing ideophones may be explicable as the result of a kind of abduction, or the perfection of a conditional. That is: If we admit that all performances tend to be repeated, then all ideophones will be. But if all ideophones are reduplicated, then, perhaps by abduction, all reduplications can potentially be counted as ideophones. More generally, the perfection of conditionals may be only a special case of the principle of association whereby an incidental feature of a form at one stage is harnessed as its defining, in fact, generative feature at another later stage.

I conclude that the reduplication of ideophones is not an accidental property but that it should in fact be added to the universal roster of canonical characteristics that more or less serve to define them. Ideophones often reduplicate, not because they iconically signal repetition or plurality, not because they are decorative, and not because they are baby talk, but because reduplication has (as one of its many functions) that of signaling and drawing attention to a performance. People who produce them, particularly as gestural accompaniments, are essentially performers, and for most amateurs, performance or showing off is the same thing as overdoing it. A parochial note: the words ‘stagey’ and ‘histrionic’ in colloquial English are virtually synonymous with ‘exaggerated’; and what is called a ‘stage whisper’ is one that is loud enough that one can hear it from the cheap seats.

Welmers’ tongue-in-cheek definition of ideophones as “those words which are such fun to use” (Reference Welmers1973:474) comes very close to this idea. They are vivid, but they are also ludic, insofar as we know that their producers, like all short-term mimics and performers, are play-actors ‘just having fun’ pretending, until they leave the stage.

The conceptual overlap between ideophones and manner adverbs has already been noted. Not surprisingly, manner adverbs are themselves often produced by reduplication of a nominal or adjectival stem. This is the case in many languages, among them Spanish, Italian, Karen, Akoose (Hedinger Reference Hedinger2008:184, 248), Koromfe (Rennison Reference Rennison1997:71, 471), Cambodian (see the discussion in Chapter 2), and Turkish (Lewis Reference Lewis1967:193). In fact, some of these clearly derived adverbs in Turkish seem to be nothing other than ideophones themselves:

Now, given that ‘almost any adjective may modify a verb’, there are intriguing minimal contrast pairs in Turkish like:

‘Go slowly’, where the bare adjective modifies a verb, on the one hand, and

‘We were walking slowly’, where the adjective is reduplicated.

Is the second more of a performance, more likely to elicit an accompanying gesture or pantomime, than the first?

In Turkish, as sporadically in Romance (cf. Spanish despacio despacio ‘slowly’, Italian diritto diritto ‘directly, in a straight line’, some manner adverbs are simply reduplications all the time: ayrɪayrɪ separately’, sɪk sɪk ‘often’, refte refte ‘gradually’.

3.1.2 Ideophones Deriving from other Word Classes, Part Two: By Sound Symbolism

The number of ideophones in languages like Japanese (2000+) and Korean (4000+) is so large that there must be some systematic means for producing more of them. Lee (Reference Lee1992) not only elevates sound symbolism to a definitional criterion for ideophones in Korean: She opens the possibility that all words are potentially vulnerable to sound symbolic deformations (compare Gordon and Heath (Reference Gordon and Heath1998), who argue on the basis of sound changes initiated by women that this is the case in English, perhaps even universally), and that once such deformation has occurred, the derived word in question is ipso facto an ideophone. That is, if little, tiny are prosaic words in English, the jocular mispronunciations leetle, teeny may be ideophones.

Given that such sound symbolism is almost invariably iconically motivated (for high front vowels = ‘small size’ over a sample of 136 languages, cf. Ultan (Reference Ultan and Greenberg1978)), it can be identified as fundamentally a dramatic performance, in accordance with the histrionic origins hypothesis.

3.1.3 Ideophones Deriving from other Word Classes, Part Three: Via Histrionic Production

In principle, all language may be performed histrionically (cf. Bolinger Reference Bolinger1986), with elaborate paralinguistic cues of the speaker’s state of mind, and thus partake of properties that are typically the preserve of ideophones. At the risk of draining my definition of empirical content entirely (but entirely in the spirit of the gestural origins hypothesis), I would like to suggest that all ordinary language, when it is so accompanied by gesture, has the potential for being ideophonic.

We have noted that histrionic manners of speaking accompany ideophones. Shona is no exception to this pattern. But Fortune goes a step beyond this when he suggests that “extremes of pitch are part of the realization of some ideophones” (Reference Fortune1962:31), which I take to mean that even pronouncing any prosaic word in a histrionic manner can make it an ideophone. Examples offered by Fortune are

Unfortunately, he does not offer any information on the putative prosaic roots of these words.

This kind of production does not normally figure in grammatical descriptions, which are records of the culturally shared, ergo conventionalized, aspects of language. Thus grammars of languages like Cambodian and Japanese typically fail to note that women’s deferential speech is characterized by high pitch and often nasalization, since this is a feature that is (a) personal in its expression, and (b) so iconic that it is shared with other languages, possibly even in the vocal productions of other species (cf. Ohala Reference Ohala1983, Reference Ohala, Hinton, Nichols and Ohala1994). Only where emotional intonation has become conventionalized in a language-specific way does it warrant grammatical description, and even this not often. The following passage from Thompson’s grammar of Vietnamese is a good description of intonation being used to convey emotional attitudes:

The high-pitched repeated version of to is ideophonic, as is also the fact of repetition itself, perhaps.

3.2.1 Ideophones Deriving from other Word Classes by other Affixational Means

In Kisi, nouns can be a source of ideophones (albeit only when reduplicated (?)).

In Shona, some verbs can become ideophones through addition of a suffix –(yi):

Or through the loss of a final syllable:

In Bantu in general (Samarin cites Lonkundo, Lamba, Mwera, Tswana, Zulu, Reference Samarin1971:142), ideophones are marked by partial reduplication, but also by the loss of personal and infinitival inflections and the addition of a specific de-verbalizing (ideophone-forming) suffix, usually -ɛ. Thus in Mwera:

In Zulu, Samarin cites Doke (Reference Doke1927:270) on an ideophone-forming suffix –iyani, which, added to a normal verb too, indicates ‘a happy state of mind’:

In Hausa, ideophones are phonologically distinct from prosaic words in allowing consonant finals (Newman Reference Newman, Voeltz and Kilian-Hatz2001:242). They may also therefore (?) be derived from corresponding verbs by the deletion of their final vowel or even by the reanalysis of a final glide as a consonant, rather than a vowel (Newman Reference Newman, Voeltz and Kilian-Hatz2001:255, 243). So fat ‘white’ can only be an ideophone (compare the prosaic word farī). But zaw ‘very sweet’ behaves like a CVC word rather than a CVV word.

3.2.2 Ideophones from Intensives

A number of ideophones may have originated simply as intensifiers on conventional adjectives or adverbs, and can be thought of as arriviste members of the category. I would include under this rubric all or most of the necessarily non-iconic ideophones for color, many of which occur (typically or exclusively) as modifiers of the conventional adjectives and play the same role as intensifiers do (Hutchison 1981: passim; Lydall Reference Lydall and Yimam2000:890; Moshi Reference Moshi2008:199–200). Examples are

Lithuanian offers another possible example. The originally infinitival suffix –ti signals intensivity, and it may be this function that is exemplified in

but it could also be functioning ipso facto as an ideophone (Wälchli Reference Wälchli and Arkadiev2015).

If the most frequent and specific meaning of ideophones when acting as adverbs of manner is to convey intensity (Samarin Reference Samarin1971:157, citing Werner 1919:194–5), then by abduction, perhaps any adverbial intensifier may be perceived as an ideophone. This perception may be reinforced in cases where the intensifier is itself reduplicated, as it almost always is in Jamsay:

As for the sources of intensives themselves, it could be that many of them derive etymologically from synonyms. There is a tendency in English and other languages to substitute a pair of synonyms for a single word: consider the common airport announcement:

This conscription of former synonyms seems to be at work in at least one language, Khmer, where intensifiers can often still be traced back to erstwhile synonyms (Haiman Reference Haiman2011:117–19). Samarin cites the ideophone (lɛmbɛ)² ‘soft’, which occurs as an (ideophonic) intensifier of kɔ ‘soft’ in Bobangui, but notes that this ideophone functions as the prosaic word meaning ‘soft’ in other Bantu languages (1971:149). Heath gives péy péy as the intensifier for dògó ‘finished’ in Jamsay, but notes that it also means ‘exhausted, depleted’ (Heath Reference Heath2008:248).

3.2.3 Words Conscripted to Act as Ideophones by Virtue of Their Novelty

Nonce formations are by definition novel. This leads us finally to another unsystematic source of ideophones with illegitimate pedigrees. By the same abductive reasoning that conscripts cusswords, reduplications, and intensives as ideophones, novel words arising via any means whatsoever may also be so conscripted. A borrowing from an unknown language may be novel. It is for this reason, I believe, that -lɛmbɛ, a word meaning ‘soft’ in other Bantu languages, but possibly unfamiliar to speakers of Bobangui, can be employed as an ideophone to accompany the prosaic word kɔ ‘soft’. The same kind of reasoning is at work in random examples like the use of dipiiii ‘a very long time’ as an ideophone in Sango (Samarin Reference Samarin, Voeltz and Kilian-Hatz2001:325). The source, French depuis, is entirely prosaic. Southern Soso has conscripted the English numeral six as an ideophone meaning ‘running’ (Kunene Reference Kunene1978:37: the conventional pantomimed gesture for running, waggling of the right thumb above a clenched fist, happens to be homo-‘phon’- ous with the sign for the numeral six.)

Looking ahead somewhat, we may ask: If novelty characterizes an ideophone, what happens after their novelty fades? In those cases where ideophonic status is conferred by novelty alone, as these few random examples would suggest, we might predict that ideophones may become conventional words in a rather short time. For this reason, we might not expect to find ideophones in current languages having roots that can be traced back to ancient sources. In fact, there will be evidence within familiar languages to support this conjecture (cf. Chapters 4 and 6).

3.3.1 An Urgent Prospect of Falsifiability

While I stated at the outset of this book that my entire proposal, like all other conjectures about the origin of language, is, alas, unfalsifiable, this was perhaps not entirely true. Much depends on the assumption that ideophones are if not always entirely nonce creations, then at least ephemeral spontaneous and opportunistic discoveries. This assumption underlies Paul’s practice in identifying as cases of original creation words that have no etymologies within Germanic, and it is compatible with the impression of Fortune and others that such words, unlike conventional prosaic vocabulary items, do not have reconstructable histories in Bantu and other language families. Now Derek Fivaz has, as noted, made the claim that some Bantu ideophones can be traced to older roots, but the three languages for which he has reconstructed such ideophones are contiguous, which raises the possibility that their cross-linguistic similarity is due to mutual borrowing rather than common origin. And Marianne Mithun, writing about apparent Iroquoian cognates, has made the very important point that their phonetic identity suggests borrowing rather than common origin as well.

Overall, reliable cognates and reconstructions from language families that are still virtually unknown as families to Western scholarship are lacking. And these are lacking because most of the extant languages for which ideophones have been richly attested belong to families whose other members have received no descriptions at all. Take, for example, Siwu, also known as Akpafu (Heine Reference Heine1968:41). This is, of course, one of our most important sources, thanks to Dingemanse’s exemplary monograph – but it is also one of fourteen Central Togo (Dakubu and Ford Reference Dakupu, Ford and Dakupu1988) or Togo remnant languages (Heine Reference Heine1968) that belong to the Kwa family, and ultimately are members of the Niger-Congo group. Almost everything we know about the other thirteen members of the Togo remnant family (Basila, Balemi, Logba, Adele, Likpe, Santrokofi, Avatime, Nyangbo-Tafi, Bowili, Ahlo, Kposo, Kebu, and Animere) we owe to the pioneering account in Heine (Reference Heine1968). And this account deals with nothing but sound correspondences and nominal morphology. There is nothing in this invaluable work even about verbal morphology, to say nothing of outcasts like ideophones. There are no grammatical descriptions remotely comparable to Dingemanse’s for any of the other languages of the Central Togo group.

Now it happens that these languages, although spoken in a small geographical area, are not contiguous, and “are very different from each other” (Dakubu and Ford Reference Dakupu, Ford and Dakupu1988:120). Moreover, “in general, the communities speaking Central Togo languages are not highly bilingual in each other’s languages” and “no CT language functions as a lingua franca between CT communities” (Dakubu and Ford Reference Dakupu, Ford and Dakupu1988:124). If it could be demonstrated that the ideophones so richly attested in Siwu shared cognates in some or most of the other Central Togo languages, then this would in my opinion severely, perhaps irreparably, weaken my hypothesis that ideophones are in any significant etymological respect a different kind of part of speech from nouns, verbs, and the other prosaic vocabulary. These other languages await investigation, though not for very long: The communities of speakers are very small (Kposo is the largest, with 82,000 speakers, Adele and Logba the smallest, with not more than 2000, cf. Dakubu and Ford (Reference Dakupu, Ford and Dakupu1988:120–1)) and almost certainly diminishing. Comparable challenges await in other language families.

I will turn in the next chapter to those ideophones that may have arisen not as onomatopes, acts of acoustic mimesis, but as vocal charades, or acts of proprioceptive, articulatory mimesis. It is noteworthy that these candidates, far from being accompanied by manual or other gestures, were themselves performed as vocal gestures in their own right.